JP4859456B2 - Data schema mapping program and computer system - Google Patents

Description

  The present invention relates to a technique for associating data item names among a plurality of data schemas.

  In recent years, the need for Internet-based BtoB (business-to-business) services has increased, and international standardization and industry standardization to improve the interoperability of information and services between businesses has been progressing. For example, for EDI (electronic data exchange), the industry standard XML / EDI of the ebXML (electronic business XML) specification has become mainstream, and for financial information, XBRL (eXtensible Business Reporting Language) has become mainstream. It is necessary to associate data items described in company-specific specifications with standard specification items. There are several tools for mapping between various original format files and EDI standard format files as mapping tools, but all use GUI to define the correspondence of data items between multiple database formats. It is necessary to use it manually. As a result, it takes time to deal with changes such as the introduction of new database systems and information retrieval systems and version upgrades to industry standards, and problems such as the accumulation of know-how for mapping item names have occurred personally. ing.

  As a conventional technique related to the association of the data item names, there is known a method using a dictionary or ontology (relationship of vocabulary used for specification description) prepared in advance. For example, in Patent Document 1, a data item name is decomposed into “qualifier + main word + classified word” using a dictionary, and the similarity of the data item name is calculated based on whether each element matches the dictionary. A method is disclosed.

  Patent Document 2 discloses a method for generating an approximate search condition using an ontology by describing a hierarchical relationship or a synonym relationship between concepts in an ontology as information retrieval means.

Patent Document 3 discloses a technique for taking in an electronic document and outputting data indicating a conceptual relationship between two morphemes.
JP-A-8-249338 JP 2003-345821 A JP 2005-157823 A

  In the above prior art, it is necessary to describe hierarchical relationships and synonymous relationships between concepts in advance in a dictionary or ontology for associating data item names. In addition, in the association of data item names, it is necessary to select corresponding ones from data item names composed of similar component words, and thus the conventional method of calculating “similarity” cannot narrow down candidates. There is a problem. For example, when associating which data item name “order date” in one data schema corresponds to which data item name in another data schema, the “order date” and “order number” are based only on the similarity. Can be obtained, but it is difficult to narrow down.

  Therefore, the present invention has been made in view of the above problems, and when linking item names between different data schemas, vocabularies that are related but should be distinguished are extracted, and a meaningful discrimination relationship is established. This is intended to narrow down the candidates for data item name association with high accuracy.

  The present invention refers to the first data schema describing the data structure and the second data schema, and the names of the data items constituting the first data schema and the data item names constituting the second data schema. A program for associating, extracting data item names constituting the first data schema, extracting data item names constituting the second data schema, and extracting the first data schema and An element concept included in each data item name of the second data schema is extracted, a discrimination relationship between the data item names is set from the extracted element concept, and the data item name of the first data schema and the second Each data item name is read from each data schema, and a correspondence relation is set for the two read data item names based on the discrimination relationship. To store the correspondence between the data item name.

  The element concept is extracted by reading two data item names for each of the first data schema and the second data schema and comparing the character strings of the two data item names to obtain synonymous character strings. When included, a common set of character strings out of character strings constituting the two data item names is extracted as a first element concept, and the character strings having the same meaning are compared by comparing the character strings of the two data item names. When a column is included, a set of character strings excluding a common character string from among character strings constituting the two data item names is extracted as a second element concept.

  Therefore, according to the present invention, by using the discrimination relationship and temporal order relationship between the vocabulary concepts constituting the data item name, the data item name correspondence candidates can be narrowed down. For example, based on the knowledge that ““ number ”and“ date ”are distinguishable concepts (concepts to be distinguished),” “order number” among the candidate data item names corresponding to “order date” Can be determined to be excluded from the candidates because “order date” is a distinctive concept. In other words, it is possible to narrow down item names with high accuracy by adding a discrimination relationship to a concept that can be treated as an item having a different meaning when a compound word is used.

  In addition, by using the temporal order relationship of the item names, it is possible to select a candidate for the correspondence relationship of the data item names with higher accuracy.

  Hereinafter, an embodiment of the present invention will be described with reference to the accompanying drawings.

  In the first embodiment, in order to associate an item name of a target database with an item name of a source database, a discriminating relationship between concepts (synonymous with a concept to be distinguished or Data schema mapping definition support system that has a function to extract the relations that cannot be similar) and temporal order relations and store them in the ontology, and a function to narrow down the correspondence of the data item names using the ontology An example is shown.

  FIG. 1 shows the first embodiment, and supports the creation of a “data schema mapping definition file” indicating the correspondence of data items between different data schemas for data linkage within a company or between companies. The block diagram of a computer system is shown.

  The computer system includes a CPU 101 that performs arithmetic processing, an input device 102 that includes a keyboard, a mouse, and the like, a display device 103 that displays arithmetic results and the like, a storage device 110 that stores data and programs, and temporarily. And a memory 108 for storing data (table or the like).

  The storage device 110 includes an input data storage unit 104 that stores input data, an ontology construction data storage unit 105 that will be described later, and an ontology storage unit that stores an ontology that defines vocabulary relationships used in specification descriptions. 106 and a data schema mapping definition file storage unit 107 for storing the generated data schema mapping definition.

  The input data storage unit 104 includes a source data schema definition document storage unit 1041, a source instance document storage unit 1042, a target data schema definition document storage unit 1043, and a target instance document storage unit 1044.

  The source data schema definition document storage unit 1041 stores a document that defines a schema of a source database. The source instance document storage unit 1042 stores actual data of a reference database.

The target data schema definition document storage unit 1043 stores a document that defines a schema of a target database that associates an item name with a source database. The target instance document storage unit 1044 stores actual data of the target database. In addition,
2A to 2D show the source data schema definition document storage unit 1041, the source instance document storage unit 1042, the target data schema definition document storage unit 1043, and the target instance document storage unit 1044 of the input data storage unit 104. A description example of the stored document is shown. These documents are described in XML / EDI of the above-described ebXML (electronic business XML) specification, XBRL (eXtensible Business Reporting Language) or the like in the case of financial information.

  FIG. 2A shows an example of the source data schema definition document 10411, where “order date” is defined as a data item name (element name) with a date data type, and “transaction number” is an integer as a data item name. An example defined with a data type of

  FIG. 2B shows an example of the source data instance document 10421, in which “20050923” is stored as “order date” data and “00020033” is stored as “transaction number” data.

  FIG. 2C shows an example of the target data schema definition document 10431, in which “order date” is defined as the data item name (element name) with the date data type, and “order number” as the data item name. Here is an example where is defined with an integer data type.

  FIG. 2D shows an example of the target data instance document 10441, in which “20044051” is stored as “order date” data and “000101123” is stored as “order number” data. .

  Next, the ontology construction data storage unit 105 includes a source data schema data item information storage unit 1051, a target data schema data item information storage unit 1052, and a concept information storage unit 1053. An ontology 1061 is stored in the ontology storage unit 106.

  3A and 3B show the data structures of the source data schema data item relationship storage unit 1051 and the target data schema data item relationship storage unit 1052.

  The source data schema data item information storage unit 1051 in FIG. 3A stores data extracted from the source data schema definition document 10411 and the source data instance document 10421 shown in FIGS. 2A and 2B as described later. Store. The source data schema data item information storage unit 1051 has, in one entry, a data item name 10511 for storing the item name of the source data schema definition document, a schema name 10512 given to the source data schema definition document, and a data item name. A parent item name 10513 for storing a data item name which is a parent (superior concept) of 10511, a data attribute 10514 for storing a data type defined in the data item name 10511, and instances corresponding to the data item name 10511 are arranged in order of frequency. It consists of an instance list 10515 to be stored.

  The target data schema data item information storage unit 1052 in FIG. 3B stores data extracted from the target data schema definition document 10431 and the target data instance document 10441 shown in FIGS. 2C and 2D as described later. Store. The target data schema data item information storage unit 1052 has, in one entry, a data item name 10521 for storing the item name of the target data schema definition document, a schema name 10522 given to the target data schema definition document, and a data item name. A parent item name 10523 for storing a data item name which is a parent (superordinate concept) of 10521, a data attribute 10524 for storing a data type defined in the data item name 10521, and an instance corresponding to the data item name 10521 are stored. An instance list 10525 is included.

  The concept information storage unit 1053 stores the concepts constituting the data item names of the source data schema definition document storage unit 1041 and the target data schema definition document storage unit 1043 as described later. The ontology 1061 stores the conceptual relationship between the item names as will be described later.

  The data schema mapping definition file storage unit 107 stores a data schema mapping definition file 1071 as a result of associating data items in the source data schema definition document and the target data schema definition document.

  In the memory 108, as a program for associating data item names using the ontology 1061 describing the discrimination relationship and temporal order relationship between vocabulary concepts, a data item information capturing unit 1081, a concept extraction unit 1082, a discrimination A relationship extraction unit 1083, a synonym relationship extraction unit 1084, an order relationship extraction unit 1085, an ontology construction support unit 1086, and a data item mapping definition support unit 1087 are loaded and executed by the CPU 101.

(Import data item information)
The operation of each program that performs ontology construction support and data mapping (narrowing down the correspondence between data item names) loaded in the memory 108 will be described below.

  First, the data item information capturing unit 1081 reads the data schema definition document from the source data schema definition document storage unit 1041 of the input data storage unit 104, captures the data item information into the source data schema data item relation storage unit 1051, and sets the target. The data schema definition document is read from the data schema definition document storage unit 1043, and the data item information is taken into the target data schema data item relationship storage unit 1052.

  The processing procedure of the data item information fetching means 1081 will be described with reference to the flowchart of FIG.

  The data item information capturing unit 1081 captures data item information from the source data schema definition document storage unit 1041, data item name 10511, data schema name 10512, parent item name 10513, data attribute of the source data schema data item information storage unit 1051. It is stored in 10514 (S601).

  Further, the data item information fetching unit 1081 extracts the instances of the respective data items from the source instance document storage unit 1042, aggregates the instance frequencies, and stores them in the instance list 10515 of the source data schema data item information storage unit 1051 in order of frequency. (S602).

  Similarly, data item information is fetched from the target data schema definition document storage unit 1043 and stored in the data item name 10521, data schema name 10522, parent item name 10523, and data attribute 10524 of the target data schema data item information storage unit 1052 ( S603). Further, the data item information fetching means 1081 extracts the instances of the respective data items from the target instance document storage unit 1044, aggregates the instance frequencies, and stores the instances in the instance list 10525 of the target data schema data item information storage unit 1052 in order of frequency. (S604).

  Through the above processing, a database data schema definition document (hereinafter simply referred to as a source data schema) stored in the reference source data schema definition document storage unit 1041 and a target data schema to be interconverted with the source data schema Item names and frequency order instances of the database data schema definition document (hereinafter simply referred to as target data schema) stored in the definition document storage unit 1043 are the source data schema data item information storage unit 1051 and target data schema data item information. The data are stored in the storage unit 1052, respectively.

(Extraction of element concepts and concept relationships)
The concept extraction unit 1082 executes concept information extraction processing of item names of two data schemas (source data schema and target data schema) that perform mutual conversion. The concept extraction unit 1082 is executed after the instance of the item name and the frequency order are stored in the source data schema data item information storage unit 1051 and the target data schema data item information storage unit 1052 by the processing of FIG.

  The concept extraction unit 1082 extracts the concepts constituting the data item names of the source data schema and the target data schema and stores them in the concept information storage unit 1053. The data structure of the concept information storage unit 1053 is shown in FIG.

  The concept information storage unit 1053 includes a concept name 10551 indicating a concept included in the data item name in one entry, a position 10532 in the data item name where the concept name exists, and a data item name associated with the concept name 10551. The number of data item names 10533 having different names, and a data item name list 10534 for storing a list of data item names associated with the concept name 10531 are included.

  The processing procedure of the concept extraction unit 1082 will be described with reference to the flowchart of FIG.

  The concept extraction unit 1082 extracts two data item names 10511 from the source data schema data item information storage unit 1051 and compares the character strings of the data item names 10511 (S701 to S704).

As a result of the comparison in S704, when the two data item names 10511 are different, assuming that the two data item names A and B are the data item names B, the element concepts constituting the item name A and the item name B are cut out as follows ( S705).
(1) Concept 1 = Common character string from the beginning of item name A and item name B (2) Concept 2 = Common character string from the end of item name A and item name B (3) Concept 3 = From item name A Excluding the common character string (4) Concept 4 = item name B excluding the common character string Next, information on how to use the concepts 1 to 4 in the data item name A and the data item name B Registration is performed in the concept information storage unit 1053 (S706). For example, when the data item name A is “request number” and the data item name B is “request date”, the concepts 1 to 4 are as follows.
(1) Concept 1 = “Request”
(2) Concept 2 = nil (none)
(3) Concept 3 = “number”
(4) Concept 4 = “Date”
Therefore, information on how the concept 1, concept 3, and concept 4 are used in the concept name and the data item name are represented by the concept name 10551 in the concept information storage unit 1053, the position 10532 in the data item name, the number 10533 of data item name differences, The name list 10534 is stored.

  Finally, the conceptual relationship between concept 3 and concept 4 (that is, the difference between data item name A and data item name B) is extracted and registered in ontology 1061 (S707). The above processing is performed for all combinations of the data item names 10511 (S708 to S711).

  The concept extraction unit 1082 also executes the process of FIG. 7 on the target data schema data item information storage unit 1052.

  Through the above processing, element concepts constituting the data item names of the source data schema and the target data schema are extracted and stored in the concept information storage unit 1053. In addition, the difference between the data item name A and the data item name B of the source data schema and the target data schema is stored in the ontology 1061.

  In addition, the above element concept includes a first element concept in which item name A and item name B include a common character string, and a second element composed of differences obtained by removing the common character string from item name A and item name B. It is composed of concepts.

(Details of conceptual relationship extraction procedure)
In the conceptual relationship extraction step (S707) in FIG. 7 above, the above-described concepts 3 and 4 (that is, the difference between the data item name A and the data item name B) are detected by the discrimination relationship extraction unit 1083 and the order relationship extraction unit 1085. The relationship is extracted and stored in ontology 1061. The data structure of ontology 1061 is shown in FIG.

  In FIG. 5, ontology 1061 includes a concept name 10541 that stores concept 3 (a character string obtained by removing a common character string from item name A) in one entry, and a concept 4 (character obtained by removing a common character string from item name B). Column), a concept relationship 10543 storing the relationship between the two concept names 10541 and 10542, a position 10544 of the concept names 10541 and 10542 in the data item name A or the data item name B, and extraction An example 10545 for storing a list of the data item names A and B, and a confirmation flag 10546 indicating that the operator or administrator has confirmed the entry.

  Next, processing procedures of the conceptual relationship extraction unit 1085 and the discrimination relationship extraction unit 1083 will be described below with reference to the flowchart of FIG.

(1) Extraction of Order Relationship When the concept 2 (that is, the common character string from the end of the data item name A and the data item name B) is “year / month / day” (S801), the concept 3 and the concept 4 (that is, The temporal order relationship between the data item name A and the data item name B) is extracted. First, the time-series order relationship between the data item name A and the data item name B is extracted from the instance document storage unit 1042 or 1044 (S802).

  If the order of the data item name A is always ahead (S803), it is determined that the concept 3 is ahead of the concept 4 (S804). For example, if the “estimated date” is always ahead of the “shipped date”, it is determined that the “estimated” order is “first” than “shipped”. Conversely, if the data item name B is always ahead of the order (S805), it is determined that the concept 4 is “first” than the concept 3 (S806).

  The order relation extracting unit 1085 performs a data item name indicating a position on the time axis. Therefore, as a condition of the character string at the end of the data item, “date”, “year / month”, “month / year / time”, etc. can be considered in addition to “year / month / day”.

For example, when there are “estimate number”, “shipment number”, and “arrival number” as mapping destination candidates for “delivery number”, it is assumed that the following information is obtained regarding the temporal order relationship.
(1) Order relationship “estimate-shipment-arrival” (2) Order relationship “shipment-delivery” In this example, “estimation” is obtained by using information that “delivery” is later than “shipment”. “Shipment” can be removed from the synonym candidate “delivery”. As a result, the “estimate number” and the “shipment number” can be removed from the “delivery number” mapping destination candidates and can be narrowed down to the “arrival number”.

  Further, the same processing as described above may be performed on condition that the data attributes 10514 and 10524 of the data item name are “time” and “year / month / day”.

(2) Extraction of discrimination relationship When there is no order relationship between concept 3 and concept 4 (ie, the difference between item name A and item name B), discrimination relationship extraction means 1083 indicates that concept 3 and concept 4 are in a discrimination relationship. Determination is made (S807). For example, when the data item name A is “request number” and the data item name B is “request date”, the concept 3 is “number” and the concept 4 is “date”. Is determined. Even when the concept 2 is not “year / month / day” in S801, the process advances to S807 to set the discrimination relationship.

  Here, the “discrimination relationship” relationship is a vocabulary relationship that cannot be a synonym, and in particular, a relationship that can be raised as a different item when it is used as a database item name, or a compound word. The conceptual relationship that can be treated as different items.

  Information on the order relationship and the discrimination relationship between concepts extracted by the above procedure is stored in the concept name 10541, concept name 10542, concept relationship 10543, position 10544 in the data item name, and example 10545 of the ontology 1061 (S808).

(Data item mapping definition support)
The data item mapping definition support means 1087 supports the correspondence between the data item names in the source data schema and the target data schema by using the discrimination relationship and the order relationship between the concepts stored in the ontology 1061. The association result is stored in the data schema mapping definition file 1071 shown in FIG.

  In FIG. 12, a data schema mapping definition file 1071 includes a source pointer 107111 for storing a pointer indicating the position of the data item name in the source data schema data item information storage unit 1051 and a data item in the target data schema data item information storage unit 1052. A pair of target pointers 107112 for storing pointers indicating the positions of names are configured.

  The processing of the data item mapping definition support means 1087 for constructing the data schema mapping definition file 1071 will be described below with reference to the flowchart of FIG.

  As a method of associating data item names between the two data schemas of the source data schema and the target data schema, data items including common character strings such as “order date” and “order date” are candidates. It is possible to extract.

Therefore, the data item mapping definition support means 1087 takes out the data items one by one from the source data schema data item information storage unit 1051 and the target data schema data item information storage unit 1052, and compares the character strings of the data item names 10511 and 10521. (S901 to S904). When the data item name 10511 of the source data schema is the data item name A and the data item name 10521 of the target data schema is the data item name B, the element concepts constituting the data item name A and the data item name B are as follows: Cut out (S905).
(1) Concept 1 = Common character string from the beginning of item name A and item name B (2) Concept 2 = Common character string from the end of item name A and item name B (3) Concept 3 = From item name A Excluding the above common character string (4) Concept 4 = Item name B excluding the above common character string The features of this system are the distinction relationship and temporal relationship of the concepts that make up data item names A and B It is in the point which narrows down a matching candidate using the information about various order relations. That is, when there is a common character string in data item name A and data item name B, reference is made to ontology 1061, and concepts 3 and 4 (that is, the difference between item name A and item name B) are not in the “discrimination relationship”. On the condition (S906), Concept 3 and Concept 4 are set as synonym candidates (S907). The synonymous relationship between the concepts is stored in the concept name 10611, concept name 10612, concept relationship 10613, position 10614 in the data item name, and example 10615 of the ontology 1061 (S908). The process of S908 corresponds to the synonym relation extracting unit 1084 shown in FIG.

  Next, the data item mapping definition support unit 1087 adds the data item name B to the mapping destination candidate of the data item name A in the data schema mapping definition file 1071 (S909). The above processing is performed for all combinations of data item names (S910 to S913).

Similarly, the data item mapping definition support means 1087 performs constriction of mapping destination candidates using information on the temporal order relationship between the concept 3 and the concept 4 (that is, the difference between the item name A and the item name B). . For example, when there are “estimate number”, “shipment number”, and “arrival number” as mapping destination candidates for “delivery number”, as shown in FIG.
(1) Order relationship that “shipment” comes before “delivery” (2) Order relationship that “estimate” precedes “shipment” In this case, “estimate” and “shipment” are excluded from synonym candidates for “delivery” Can do. Thereby, the “estimate number” and the “shipment number” can be removed from the “delivery number” mapping destination candidates.

  As described above, the data item mapping definition support unit 1087 of this system narrows down the mapping destination candidates related to the item names of the two databases (data schemas) by the above procedure, and then displays the mapping destination candidates on the screen of the display device 103. indicate. A display example of the data item mapping definition support screen is shown in FIG.

  In FIG. 10, a data item mapping definition support screen 1001 displays a target schema display unit 10011 that displays a data item name of a target data schema to be associated, and a source schema display that displays a data item name of a reference source data schema. Part 10012, synonym concept candidate display part 10013 for displaying candidate data item names that are synonymous in two data schemas, and candidate data item names that are discriminating in two data schemas And a registration button 10015 for accepting an operator's operation. The synonym concept candidate display unit 10013 and the discrimination concept candidate display unit 10014 display synonym concept candidates and discrimination concept candidates included in the data item name.

  For the data item name displayed in the source schema display unit 10012, candidates for association (mapping) are displayed as mapping destination candidates in the target schema display unit 10011. When the user selects an appropriate candidate from the mapping destination candidates and selects (clicks) the registration button 10015 with the input device 102, the selected mapping destination candidate is stored in the data schema mapping definition file 1071.

  For example, in FIG. 10, when the data item name of the source data schema “Order Date” is displayed in the source schema display unit 10012 set in the display area on the right side of the screen, the target schema display set in the display area on the left side is displayed. The part 10011 displays “order date” as a candidate associated with “order date”. The operator may operate the registration button 10015 to approve the result narrowed down by the data item mapping definition support means 1087. A synonym for the data item name “order date” displayed in the target schema display unit 10011 is displayed in the synonym concept candidate display unit 10013 on the lower left side of the screen 1001. Further, the discrimination concept candidate display section 10014 on the lower right side of the screen 1001 displays a phrase that is a discrimination relationship to be distinguished from the vocabulary of the mapping destination candidates of the source data schema and the target data schema.

  Similarly, when the user selects an appropriate candidate from the synonym concept candidate and the discrimination concept candidate and clicks the registration button 10015, the information on the synonym concept and the discrimination concept of the ontology 1061 is updated according to the selection, and is shown in FIG. The confirmation flag 10546 is set to a value “1” indicating confirmation.

  In the above, an example in which the association (mapping) of the data item names of the two data schemas is approved by the operator's operation is shown. However, the mapping destination candidates narrowed down by the data item mapping definition support unit 1087 are displayed as the data schema mapping definition. It may be automatically registered in the file 1071.

(Ontology construction support means)
Finally, the ontology construction support means 1086 will be described. In this embodiment, the ontology 1061 is partially constructed in the process of the data item mapping definition support means 1087 described above. The ontology construction support means 1086 provides means for confirming the discrimination concept and the synonym concept for each concept. provide. An example of the ontology construction support screen is shown in FIG. The ontology construction support screen 901 displays a concept display unit 9011 that displays a word corresponding to the selected concept, and a concept that is synonymous with a concept that is a discrimination relationship with respect to the word displayed on the concept display unit 9011. A concept relationship display unit 9012, a concept relationship candidate display unit 9013 for displaying a phrase that is a discrimination relationship with respect to the concept displayed in the concept relationship display unit 9012, and a phrase that is a synonym relationship, and an operation of the operator A registration button 9014 is provided. The concept relationship display portion 9012 displays synonymous concepts, discrimination concepts, and temporal order relationships. The conceptual relationship candidate display unit 9013 displays synonymous concept candidates, discrimination concept candidates, and order relationship candidates stored in the conceptual relationship storage unit 1054. When the user selects an appropriate one from the concept relationship candidates and clicks the registration button 1094, the selected synonym concept candidate, discrimination concept candidate, and order relationship candidate are stored in the ontology 1061 and confirmed in the confirmation flag 10616. A value “1” indicating that the value is set is set.

  The ontology 1061 may also be constructed by automatically selecting and registering a high-priority discrimination concept candidate and a synonym concept candidate.

(Data schema mapping definition file)
The data schema mapping definition file 1071 created by the data schema mapping definition support system can integrate the database defined by the target data schema and the database defined by the source data schema.

  For example, when searching for “order date” in a database defined in the target data schema, the database system (not shown) refers to the data schema mapping definition file 1071 and the data item name of the target data schema is “ Refer to the data item name of the source data schema corresponding to “order date”. A pointer of the data item name of the source data schema is obtained from the data schema mapping definition file 1071 shown in FIG. 12, and “order date” is obtained from the source data schema data item information storage unit 1051. As a result, the database system can operate the two databases virtually integrated by performing a similar search on the “order date” with respect to the database defined in the source data schema.

  Further, when the database defined in the source data schema is integrated into the database defined in the target data schema, data may be transferred between the data item names associated in the data schema mapping definition file 1071. For example, as shown in FIG. 10, the data (instance) of the data item name “order date” defined in the source data schema is changed to the data item name “order year” defined in the target data schema data item name. What is necessary is just to write as data (instance) of "month and day".

(Summary)
As described above, in the present invention, the discrimination relationship between concepts is described in an ontology in order to narrow down data item name association candidates among a plurality of data schemas. Moreover, in order to automate ontology construction, a concept discrimination relationship is extracted from the relationship between data item names in the data schema. In other words, using the property that “a data item with the same parent is given a distinguishable name that can be distinguished as a different item”, the discrimination relationship between concepts can be determined from the terms of data items with the same parent. Extract. For example, when the data item names “request number” and “request date” are in a sibling relationship, ““ request number ”and“ request date ”are distinct concepts (concepts to be distinguished)”. Extract knowledge. Furthermore, knowledge that “the“ number ”and the“ date ”representing the difference between the“ request number ”and the“ request date ”are distinguishable concepts” ”is extracted.

  And in this invention, in order to narrow down the correspondence candidate of a data item name, the temporal order relationship between concepts is described ontology. Also, in order to automate ontology construction, temporal order relations are extracted from the instance document. For example, in the instance document, since the “shipment date” is always earlier than the “arrival date”, the temporal order relationship ““ shipment ”is earlier than“ arrival ”” is extracted.

  As described above, according to the present invention, by associating data item names between a plurality of data schemas using an ontology that describes the discrimination relationship and temporal order relationship between concepts, data item name association candidates Can be accurately and easily squeezed. In particular, as in the prior art, it is possible to automate the association of data item names that relied on manual work. As a result, it is possible to quickly respond to changes such as integration between different databases (virtual integration or physical integration), new introduction or update of databases, or database upgrades corresponding to industry standards.

<Second Embodiment>
(Information retrieval system using discrimination)
In the first embodiment, the method of extracting the conceptual relationship “discrimination relationship” from the data schema and using it for associating the data item names between the data schemas has been described. Since the “discrimination relationship” is effective when narrowing down the correspondence between the data item names of the two data schemas, it can also be applied to conversion of search conditions for searching for appropriate information from a plurality of information servers. In information retrieval, a technique called question expansion is known as means for reducing search omissions.

  Question expansion is to automatically add a word related to a word in the search question to the search question. For example, when the search question is “automobile”, “car”, “passenger car”, “private car”, and the like are added. The words to be added in the question expansion are different expressions (computer → computer), synonyms (computer → computer), broader terms (computer → computer), narrower terms (computer → computer), and the like.

  As a method of finding a word to be added, a method of creating a dictionary or a thesaurus in advance is common. For example, Patent Document 2 discloses a method of converting into approximate search conditions using an ontology. If the search condition is expanded by the measure of “similarity”, search omissions are reduced, but noise is likely to occur. Therefore, in the second embodiment, it is possible to convert candidates into more appropriate search conditions by narrowing down candidates to conditions using knowledge of “discrimination relation”. For example, in addition to adding a broader term such as “passenger car” or “car” to the search query “private car”, using the knowledge that “private car” and “commercial” have a discrimination relationship, "Can be eliminated to reduce noise.

  Hereinafter, according to the flowchart of FIG. 13, a search procedure of the information search system using the information on the hierarchical relationship and the discrimination relationship described in the ontology will be described. The flowchart in FIG. 13 is executed by the same computer as in FIG. This search system refers to a product system of a plurality of companies, and is a system for searching for product information that the user wants to search for. The system is premised on maintaining a dictionary that records conceptual relationships between words and a dictionary that records discrimination relationships between words according to the present invention. First, when the user inputs “welfare vehicle” as a product category name to be searched for (S1301), the system uses, as reference information, an example of a product part system related to “welfare vehicle”, for example, Company A as shown in FIG. The product classification system is displayed on the screen (S1302). The display format is not limited to that shown in FIG. 14, and a word related to the input word may be specified. When the user selects “elevating seat car” as the product classification name, the system recognizes “elevating seat car” as a search keyword (S1303). Further, the superordinate concept of “elevating seat car”, that is, “nursing care vehicle” and “welfare vehicle” are added to the search keyword (S1304).

  Next, with reference to information regarding the discrimination relationship as shown in FIG. 14B, the discrimination concept of the constituent words of the search keyword is set as an excluded keyword (S1305). For example, the discriminating concept of “lifting seat car” “wheelchair moving vehicle”, “stretcher moving vehicle”, the discriminating concept of “elevating” “rotation”, and the discriminating concept of “care type” “self-operated”, “public transportation "As a negative keyword. After the search keyword is expanded and the exclusion keyword is set in this way, product information that can be used for the “lifting seat car” is searched for the product systems of a plurality of companies (S1306). For example, “classified as“ lifting seat car ”or“ nursing care vehicle ”or“ welfare vehicle ”, and“ wheelchair moving vehicle ”“ stretcher moving vehicle ”“ rotation ”“ self-operated ” “None of public transportation” is searched. With this method, as shown in FIGS. 15 (a) and 15 (b), even if company B does not have a product classification name “elevating seat vehicle”, “self-operated specification vehicle” among “welfare vehicles” “Rotating seat specification vehicle” is excluded from “wheelchair moving vehicle”, and only “lift-up seat specification vehicle” is displayed as a search result (S1307). If the discrimination relationship in the present invention is not used, the search keyword can be expanded to the upper concept “welfare vehicle”, but “self-operated specification vehicle”, “rotating seat specification vehicle” and “wheelchair moving vehicle” are candidates. Cannot be excluded.

  As described above, according to the present invention, by narrowing down the search condition using the knowledge of the discrimination relationship, it is possible to convert to a more appropriate search condition, and it is possible to obtain a search result with less noise while preventing a search omission.

  The present invention can be applied to integration of different databases, and in particular, can be applied to virtual integration of databases for realizing information cooperation and service cooperation within a new company / between companies. The present invention can also be applied to an information search system that determines the concept of a phrase.

1 is a system configuration diagram of a data schema mapping definition support system according to an embodiment of the present invention. FIG. FIG. 2 is an explanatory diagram showing an example of a schema document, where (a) shows a source data schema definition document, (b) shows a source instance document, (c) shows a target data schema definition document, and (d) shows a target instance. It is explanatory drawing which shows the example of a description of a document. It is explanatory drawing which shows the data structure of a source data schema data item information storage part and a target data schema data item information storage part, (a) shows a source data schema data item information storage part, (b) Target data schema data item information Indicates a storage unit. It is explanatory drawing which shows the data structure of a conceptual information storage part. It is explanatory drawing which shows the data structure of ontology. It is a flowchart which shows an example of a data item information taking-in process. It is a flowchart which shows an example of a concept extraction process. It is a flowchart which shows an example of a conceptual relationship extraction process. It is a flowchart which shows an example of a data item mapping definition assistance process. It is a screen image which shows a data item mapping definition assistance screen. It is a screen image which shows an ontology construction support screen. It is explanatory drawing which shows the data structure of a data schema mapping definition storage part. It is a flowchart of the process which shows an example of an information search system. It is explanatory drawing which shows an example of the hierarchical relationship between words, and a discrimination relationship, (a) shows the hierarchical relationship of the word in A company, (b) shows the discrimination relationship between the words in A company. It is explanatory drawing which shows the other example of the hierarchical relationship between words, and a discrimination relationship, (a) shows the hierarchical relationship of the word in B company, (b) shows the discrimination relationship between the words in B company.

Explanation of symbols

101 CPU
102 Input Device 103 Display Device 104 Input Data Storage Unit 105 Ontology Construction Data Storage Unit 106 Ontology Storage Unit 107 Data Schema Mapping Definition Storage Unit 108 Memory 108
1041 Source data schema definition document storage unit 1042 Source instance document storage unit 1043 Target data schema definition document storage unit 1044 Target instance document storage unit 1051 Source data schema data item information storage unit 1052 Target data schema data item information storage unit 1053 Concept information storage Unit 1061 Ontology 1071 Data schema mapping definition file 1081 Data item information capturing means 1082 Concept extraction means 1083 Discrimination relation extraction means 1084 Order relation extraction means 1085 Synonym relation extraction means 1086 Ontology construction support means 1087 Data item mapping definition support means

Claims (11)

Referring to the first data schema and the second data schema describing the data structure, the data item names constituting the first data schema are associated with the data item names constituting the second data schema. A program,
A procedure of extracting data item names constituting the first data schema;
A procedure for extracting data item names constituting the second data schema;
A procedure for extracting an element concept included in each data item name of the extracted first data schema and second data schema;
A procedure for setting a discrimination relationship between data item names from the extracted element concept;
A procedure for reading data item names from the data item name of the first data schema and a second data schema, respectively, and setting a correspondence relationship for the two read data item names based on the discrimination relationship;
A procedure for storing the correspondence of the data item names;
A program characterized by causing a computer to execute. The procedure for extracting the element concept is as follows:
A procedure for reading two data item names for each of the first data schema and the second data schema;
When the character strings of the two data item names are compared to include a synonymous character string, a common set of character strings is extracted as a first element concept among the character strings constituting the two data item names And the steps to
When the character strings of the two data item names are compared and the synonymous character strings are included, a set of character strings excluding a common character string from among the character strings constituting the two data item names is a second character string. To extract as an element concept of
The program according to claim 1, comprising: The procedure for setting the discrimination relationship is as follows:
A procedure for determining a time-series order relationship for two item names corresponding to the second element concept;
When the second element concept does not include a time-series order relationship, a procedure for setting that a set of character strings constituting the second element concept is a discrimination relationship;
The program according to claim 2, comprising: The procedure for the association is as follows:
A comparison is made between the character strings of the two read data item names, and the two data item names are set in correspondence when they contain synonymous character strings and are not in a discrimination relationship. The program according to claim 1. Referring to the first data schema and the second data schema describing the data structure, generating a discrimination relationship between the data item names constituting the first data schema and the data item names constituting the second data schema A program to
A procedure of extracting data item names constituting the first data schema;
A procedure for extracting data item names constituting the second data schema;
A procedure for reading two data item names for each of the first data schema and the second data schema;
When the character strings of the two data item names are compared and the synonymous character strings are included, a set of character strings excluding a common character string from among the character strings constituting the two data item names is a second character string. To extract as an element concept of
A procedure for determining a time-series order relationship for two item names corresponding to the second element concept;
When the second element concept does not include a time-series order relationship, a procedure for setting that a set of character strings constituting the second element concept is a discrimination relationship;
The program characterized by including. The procedure for setting the discrimination relationship is as follows:
6. The program according to claim 5, further comprising a procedure for storing in the ontology a value indicating that the set of character strings and a conceptual relationship between the set of character strings is a discrimination relationship. When the character strings of the two data item names are compared to include a synonymous character string, a common set of character strings is extracted as a first element concept among the character strings constituting the two data item names And the steps to
A procedure for storing in the ontology a value indicating that the set of character strings and the conceptual relationship of the one set of character strings are synonymous;
The program according to claim 6, further comprising: A first data schema storage unit for storing the input first data schema;
A second data schema storage unit for storing the input second data schema;
Data item names constituting the first data schema are extracted from the first data schema storage unit and stored in the first data item storage unit, and data constituting the second data schema is obtained from the second data schema storage unit A data item name capturing unit for extracting the item name and storing it in the second data item storage unit;
A concept extraction unit for extracting an element concept included in each data item name of the first data item storage unit and the second data item storage unit;
A discrimination relationship extracting unit that extracts a discrimination relationship between data item names from the extracted element concept and stores it in an ontology; and
A synonym relation extracting unit that extracts a synonym relation between data item names from the extracted element concept and stores it in an ontology; and
Data item names are respectively read from the first data item storage unit and the second data item storage unit, and correspondences between the two data item names read are determined based on the ontology, and the correspondences are mapped. Mapping definition part to be stored in the file,
A computer system characterized by comprising: The concept extraction unit
A comparison unit that reads two data item names from the first data item storage unit or the second data item storage unit and compares character strings of the two data item names;
When the two data item names include synonymous character strings, a common set of character strings is extracted as a first element concept among the character strings constituting the two data item names, and the two data An element concept extraction unit that extracts, as a second element concept, a set of character strings excluding a common character string from character strings constituting the two data item names when the item name includes a synonymous character string When,
The computer system according to claim 8, comprising: The discrimination relationship extraction unit
A procedure order relation extraction unit that extracts a time-series order relation for two item names corresponding to the second element concept;
When the second element concept does not include a time-series order relationship, a discrimination relationship determination unit that determines that a set of character strings constituting the second element concept is a discrimination relationship;
The computer system according to claim 9, comprising: A search program that performs a search for an input word,
A procedure to accept input of a word as a search key,
A procedure for adding a word that is a superordinate concept of the input word as a search key based on a dictionary that records a conceptual relationship between words;
A procedure for setting each word of the search key and a word having a discrimination relationship with a component of the word as an exclusion key based on a dictionary that records a discrimination relationship between words;
Performing a search using the search key;
A procedure for outputting a search result by excluding those including the exclusion key from the search result;
A search program characterized by causing a computer to execute.

Images